faimlab/fabert_formality_classifier

Fine-Tuned FaBERT Model for Formality Classification

This repository contains a fine-tuned version of FABERT, a pre-trained language model designed for formality classification. This model has been specifically trained to classify text as formal or informal, making it ideal for applications in content moderation, social media monitoring, and customer support automation.

Model Overview

• Architecture: Built on the FABERT model, a transformer-based architecture optimized for NLP tasks.
• Task: Formality Classification – distinguishing between formal and informal language in text.
• Fine-Tuning: The model has been fine-tuned on a custom dataset containing a variety of formal and informal text.

Key Features

• Multilingual Support: This model is capable of classifying text in multiple languages, ensuring robustness in diverse linguistic contexts.
• High Performance: Fine-tuned to provide accurate predictions for formal vs. informal text classification.
• Efficient for Deployment: Optimized for real-time use in environments like social media platforms, content moderation tools, and communication systems.

How to Use the Model

You can use this model in your Python code with the Hugging Face transformers library and PyTorch. The following code snippet demonstrates how to tokenize text, make predictions, and classify whether the text is formal or informal.

import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification

# Load the pre-trained tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("faimlab/fabert_formality_classifier")
model = AutoModelForSequenceClassification.from_pretrained("faimlab/fabert_formality_classifier")

# Ensure the model runs on GPU if available
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

# Example input text
input_text = "Please find attached the report for your review."

# Tokenize the input
inputs = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True, max_length=512)

# Move the model and input to GPU if available
inputs = {key: value.to(device) for key, value in inputs.items()}

# Make predictions
with torch.no_grad():
    outputs = model(**inputs)
    logits = outputs.logits

# Get the predicted label
predicted_label = logits.argmax(dim=1).item()
print(f"Predicted Label: {predicted_label}")